11:05 a.m.
Hi,
I'm trying to build an installer-cd with livecd-creator :)
So I finally succeeded to build a livecd with livecd-creator under fedora7.
When I boot this CD, I switch to the text-console, log in (into textmode) as
root, run
/etc/init.d/messagebus start
/etc/init.d/haldaemon start
and run the liveinst.sh script from anacondas CVS
anaconda/liveinst/liveinst.sh (downloaded with curl.. for now)
(I plan to put these steps into /etc/rc.d/init.d/fedora-live but for debugging
it's easier to run them manually.)
Then I configure anaconda manually (currently just using the defaults, which,
again, I later plan to do with a kickstart file) and anaconda successfully
partitions the disc and copies the system from the livecd to harddrive.
But in the end the installation fails, log is attached.
I would be happy if someone who is more familar with anaconda logs can tell me
what went wrong...
regards,
Holger
12:14 p.m.
On Thu, 2007-05-10 at 18:05 +0200, Holger Levsen wrote:
Then I configure anaconda manually (currently just using the
defaults, which,
again, I later plan to do with a kickstart file) and anaconda successfully
partitions the disc and copies the system from the livecd to harddrive.
But in the end the installation fails, log is attached.
I would be happy if someone who is more familar with anaconda logs can tell me
what went wrong...
Bug in the text mode case as really only the graphical install has been
tested for live installs. Fixed in anaconda CVS
Jeremy
12:40 p.m.
Hi Jeremy,
On Thursday 10 May 2007 19:14, Jeremy Katz wrote:
Bug in the text mode case as really only the graphical install has
been
tested for live installs. Fixed in anaconda CVS
Yay! Thanks for telling (and probably fixing) :)
Will this fix go into fedora7 or is it too late already?
regards,
Holger
12:47 p.m.
On Thu, 2007-05-10 at 19:40 +0200, Holger Levsen wrote:
On Thursday 10 May 2007 19:14, Jeremy Katz wrote:
> Bug in the text mode case as really only the graphical install has been
> tested for live installs. Fixed in anaconda CVS
Yay! Thanks for telling (and probably fixing) :)
Will this fix go into fedora7 or is it too late already?
It'll go into Fedora 7
Jeremy
2:01 p.m.
Hi,
I used the --shell option of livecd-creator and copied the fixed text.py from
the anaconda CVS into the live-system and gave it another try. And I hit
another exception :)
anacdump.txt is attached :) (well named a bit differently, but..)
regards,
Holger
9:41 a.m.
On Fri, 2007-05-11 at 21:01 +0200, Holger Levsen wrote:
I used the --shell option of livecd-creator and copied the fixed
text.py from
the anaconda CVS into the live-system and gave it another try. And I hit
another exception :)
anacdump.txt is attached :) (well named a bit differently, but..)
This looks like it actually should be fixed with the current anaconda
code (I don't see calls to pumpNetDevice and I know that dcantrell was
changing some things around there at one point). We need to just get a
new build done today
Jeremy
11:44 a.m.
New subject: building an automatic installer with livecd-creator
Hi, Jeremy,
how would you estimate the chances to successfully backport
current livecd-tools/anaconda from FC7 to FC5?
Thanks in advance.
Regards,
Vladimir
Jeremy Katz пишет:
On Fri, 2007-05-11 at 21:01 +0200, Holger Levsen wrote:
> I used the --shell option of livecd-creator and copied the fixed text.py from
> the anaconda CVS into the live-system and gave it another try. And I hit
> another exception :)
>
> anacdump.txt is attached :) (well named a bit differently, but..)
This looks like it actually should be fixed with the current anaconda
code (I don't see calls to pumpNetDevice and I know that dcantrell was
changing some things around there at one point). We need to just get a
new build done today
Jeremy
--
Fedora-livecd-list mailing list
Fedora-livecd-list(a)redhat.com
https://www.redhat.com/mailman/listinfo/fedora-livecd-list
11:53 a.m.
New subject: building an automatic installer with livecd-creator
On Mon, 2007-05-14 at 20:44 +0400, Vladimir Shebordaev wrote:
how would you estimate the chances to successfully backport
current livecd-tools/anaconda from FC7 to FC5?
It would be a fair bit of work. And since FC5 is officially EOL'd with
the release of Fedora 7, it's not really something that I'm going to
take the time to do. I'd be willing to look at clean patches, but I
suspect that's easier said than done
Jeremy
1:07 p.m.
New subject: livecd-tools backport to FC5
Jeremy, thanks,
it seems I can't avoid trying.
I will send the patches here if they are worth looking at :)
Regards,
Vladimir
Jeremy Katz пишет:
On Mon, 2007-05-14 at 20:44 +0400, Vladimir Shebordaev wrote:
> how would you estimate the chances to successfully backport
> current livecd-tools/anaconda from FC7 to FC5?
It would be a fair bit of work. And since FC5 is officially EOL'd with
the release of Fedora 7, it's not really something that I'm going to
take the time to do. I'd be willing to look at clean patches, but I
suspect that's easier said than done
Jeremy
--
Fedora-livecd-list mailing list
Fedora-livecd-list(a)redhat.com
https://www.redhat.com/mailman/listinfo/fedora-livecd-list
11:47 a.m.
Hi,
On Monday 14 May 2007 16:41, Jeremy Katz wrote:
This looks like it actually should be fixed with the current
anaconda
code (I don't see calls to pumpNetDevice and I know that dcantrell was
changing some things around there at one point).
I was using the current anaconda.rpm from fedora7, plus text.py from CVS. I
didn't bother to build the whole rpm myself.
We need to just get a
new build done today
I'd be happy to test a new rpm before you release it :)
regards,
holger
12:39 p.m.
Hi,
I'm still working on building a fedora7 livecd which turns into a textmode
installer...
Basically I'm there, but I still have found some problems (which are all
workaroundable, but...).
I've been using fedora7 as of today and livecd from git as of today, to build
the image. The config is at
http://xs-dev.laptop.org/holger/livecd-xs-installer-9.ks and the resulting iso
is at http://xs-dev.laptop.org/holger/My-Fedora-LiveCD-9.iso
The first problem is at the very end of the build process:
anaconda-runtime not installed; not setting up mediacheck
This is strange, as a.) anaconda-runtime is installed on the build-host and
b.) is included in livecd-xs-installer-9.ks
The CD boots fine and starts anaconda in textmode. Then there is the second
problem: the questions about installing grub only appear, if I choose to
review the partioning layout! (I basically just say "ok" to all questions.)
Then, when I eject the CD and boot the installed system, it starts grub, but
there is not /boot/grub/menu.lst. When I configure grub manually (select root
device, kernel & initrd), the system boots fine.
But then I hit the fourth problem: I can login as root without a password. (In
the newly installed system.)
Attached is anaconda.log from /tmp/anaconda.log after an installation without
reviewing the partioning layout and thus without the grub questions, the
atached anacond2.log is from one with the grub questions. Except for that and
the dates, those are identical though.
regards,
Holger
1:32 p.m.
Hi,
On Monday 21 May 2007 19:39, Holger Levsen wrote:
The first problem is at the very end of the build process:
anaconda-runtime not installed; not setting up mediacheck
This is strange, as a.) anaconda-runtime is installed on the build-host and
b.) is included in livecd-xs-installer-9.ks
This is still there.
The CD boots fine and starts anaconda in textmode. Then there is the
second
problem: the questions about installing grub only appear, if I choose to
review the partioning layout! (I basically just say "ok" to all questions.)
This too.
But the third (no grub installed) and fourth (no root passwd) problem are gone
since yesterdays updates in fedora7, yay.
So now I want to use a kickstart-file with anaconda, to install the system
from the livecd automatically. And hit another exception, anaconda-dump file
attached.
The kickstart-file is part of the livecd config-file :) and is online at
http://xs-dev.laptop.org/holger/livecd-xs-installer-14.ks
anaconda is executed from liveinst.sh like this:
ANACONDA="/usr/sbin/anaconda --method=livecd://$LIVE_BLOCK --lang
$LANG --kickstart=/root/anaconda-ks.cfg"
regards,
Holger
P.S.: would you prefer if I directly file bugs in bugzilla, or is it fine like
this?
1:58 p.m.
Hi,
On Thursday 24 May 2007 20:32, Holger Levsen wrote:
So now I want to use a kickstart-file with anaconda, to install the
system
from the livecd automatically. And hit another exception, anaconda-dump
file attached.
This also happens with anaconda running in X, just checked. Same exception, so
it's no textmode issue :)
regards,
Holger
The kickstart-file is part of the livecd config-file :) and is online
at
http://xs-dev.laptop.org/holger/livecd-xs-installer-14.ks
anaconda is executed from liveinst.sh like this:
ANACONDA="/usr/sbin/anaconda --method=livecd://$LIVE_BLOCK --lang
$LANG --kickstart=/root/anaconda-ks.cfg"
9:59 a.m.
Hi,
On Friday 25 May 2007 20:58, Holger Levsen wrote:
On Thursday 24 May 2007 20:32, Holger Levsen wrote:
> So now I want to use a kickstart-file with anaconda, to install the
> system from the livecd automatically. And hit another exception,
> anaconda-dump file attached.
This also happens with anaconda running in X, just checked. Same exception,
so it's no textmode issue :)
I'm aware that it's not even a week since I reported this bug, but it's quite
a blocker for us. Do you need more details or you want a bugreport to track
this?
regards,
Holger
9:59 a.m.
On Wed, 2007-05-30 at 16:59 +0200, Holger Levsen wrote:
On Friday 25 May 2007 20:58, Holger Levsen wrote:
> On Thursday 24 May 2007 20:32, Holger Levsen wrote:
> > So now I want to use a kickstart-file with anaconda, to install the
> > system from the livecd automatically. And hit another exception,
> > anaconda-dump file attached.
> This also happens with anaconda running in X, just checked. Same exception,
> so it's no textmode issue :)
I'm aware that it's not even a week since I reported this bug, but it's quite
a blocker for us. Do you need more details or you want a bugreport to track
this?
Sorry, been preoccupied with getting the Fedora 7 release finalized and
all of the i's dotted, t's crossed, etc.
But yes, a bug report against anaconda would be great. Much less easy
to get lost that way than mail :-)
Jeremy
10:58 a.m.
Hi,
On Wednesday 30 May 2007 16:59, Jeremy Katz wrote:
Sorry, been preoccupied with getting the Fedora 7 release finalized
and
all of the i's dotted, t's crossed, etc.
No problem, I understand :) Only that we need to this bug to be fixed to
deploy the OLPC schoolservers as planned ;)
But yes, a bug report against anaconda would be great. Much less
easy
to get lost that way than mail :-)
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=242083
regards,
Holger
12:19 p.m.
Hi,
On Wednesday 30 May 2007 16:59, Jeremy Katz wrote:
But yes, a bug report against anaconda would be great. Much less
easy
to get lost that way than mail :-)
which was fixed by you yesterday, yay & thanks!
But now I found yet another bug ;) :/
After copying the live image to hd and doing the post-installation changes, it
hits another exception, see attachment. The visible result is, that the
initrd is not created on the installed system, so it doesnt boot at all...
Config, build.log and iso are available at:
http://xs-dev.laptop.org/holger/200706051001-livecd-xs-installer.ks
http://xs-dev.laptop.org/holger/200706051001-build.log
http://xs-dev.laptop.org/holger/200706051001-XS.iso
I'll file a proper bugreport as soon as I have confirmed this with a livecd
built on fedora7 (which should be either tonite or tomorrow morning), the
system currently still has the pre7 from last week. livecd is latest git.
regards,
Holger
1:47 p.m.
Hi,
On Tuesday 05 June 2007 19:19, Holger Levsen wrote:
After copying the live image to hd and doing the post-installation
changes,
it hits another exception, see attachment. The visible result is, that the
initrd is not created on the installed system, so it doesnt boot at all...
It's in the config, but I should have said this directly: I used fedora7 for
the livecd, with /usr/lib/anaconda/installclass.py to fix #242083.
regards,
Holger
10:57 a.m.
On Tue, 2007-06-05 at 19:19 +0200, Holger Levsen wrote:
On Wednesday 30 May 2007 16:59, Jeremy Katz wrote:
> But yes, a bug report against anaconda would be great. Much less easy
> to get lost that way than mail :-)
which was fixed by you yesterday, yay & thanks!
But now I found yet another bug ;) :/
Hey, I recognize this traceback. I actually fixed it up a week or so in
anaconda, but it was after we finished up Fedora 7. But I think if you
hit this, you're going to have another error that may trip things up.
The fix is basically adding
class Error(EnvironmentError):
pass
right above the definition of copytree in livecd.py
Jeremy
11:12 a.m.
Hi,
On Wednesday 06 June 2007 17:57, Jeremy Katz wrote:
Hey, I recognize this traceback.
:-)
I actually fixed it up a week or so in
anaconda, but it was after we finished up Fedora 7. But I think if you
hit this, you're going to have another error that may trip things up.
So just taking livecd.py from tunk will not be enough? Or what do you mean
with another error?
The fix is basically adding
class Error(EnvironmentError):
pass
right above the definition of copytree in livecd.py
I'll just take livecd.py from trunk and report back in approx. two hours what
the result is.
regards & thanks once again,
Holger
6:36 p.m.
Hi,
On Wednesday 06 June 2007 18:12, Holger Levsen wrote:
So just taking livecd.py from tunk will not be enough? Or what do you
mean
with another error?
Indeed I hit another error, anaconda dump is attached again.
This time I used anaconda from F7 plus installclass.py from CVS trunk (to fix
#242083) and livecd.py, also from trunk.
The change you ment was from may 29th:
* livecd.py: Define Error so we don't traceback with that when
there's a problem copying (#241754)
regards,
Holger
6160
days inactive
6190
days old
livecd@lists.fedoraproject.org
21 comments
3 participants
participants (3)
-
Holger Levsen -
Jeremy Katz -
Vladimir Shebordaev